The present invention relates to an image picturing method and an image picturing device, which use an electron beam, and more specifically, to an electron beam image picturing method and image picturing device, which are effective for the correction of the proximity effect.
In the drawing of an image using an electron beam, the phenomenon called back scattering occurs, in which electron made incident on a substrate on which an image is to be written, scatters within the substrate and appears once again on the surface of the substrate. The range in which back scattering reaches (that is, the diameter of the back scattering) is dependent on the energy of incident electron and the material of the substrate.
As the incident energy becomes larger, the diameter of the back scattering becomes larger, and as the atomic number of the material of the substrate is larger, it becomes smaller. The ratio of the energy of the back scattering with respect to the incident energy is called back scattering coefficient, and the back scattering coefficient becomes larger as the atomic number of the material of the substrate is larger. However, for substrates of the same material, the coefficient may differ from one another depending upon the type of the light sensitive agent (resist) applied thereon.
The energy accumulated in a resist is the total of the energy of the incident electron and that of the electron back-scattered. In the case of a pattern having a drawing area ratio, the pattern is greatly influenced by the back scattering, and as compared with a pattern having a small drawing area ratio, the energy accumulated in the resist becomes larger for the same incident energy. The drawing area ratio used here is the ratio of the area of the image drawn to the area of desired region in the surface of the substrate.
The phenomenon in which the effective irradiation amount differs when the drawing area ratio differs within the region where the back scattering reaches, is called proximity effect. As a result of the proximity effect, patterns drawn with the same irradiation amount, will have resist pattern sizes different from each other when the drawing area ratio differs from one another. As a solution to this, the correction of the irradiation amount in accordance with the image drawing area ratio (the correction of proximity effect) is generally carried out. The correction amount can be obtained by calculation from the data of the pattern to be drawn, and the correction amount is stored as the irradiation amount correction data together with the pattern drawing data.
The irradiation amount correction data is stored usually as data of 256 gradations. As the number of gradations is larger, it is possible to make the correction of a higher efficiency; however the data amount becomes larger. Conventionally, the correction to be carried on is divided to correspond to 256 gradations at a certain width, which is equivalent to 0.7% of the reference irradiation amount for one gradation. Therefore, the correction amount of 0% to less than 0.7% is for the first gradation, the correction amount of 0.7% to less than 1.4% is for the second gradation, and so on.
As a pattern has a higher image drawing area ratio, the ratio of the back scattering energy occupied in the energy accumulated in the resist becomes larger, thus the contrast of pattern becomes poor. Therefore, the variance in the size of the resist pattern, which occurs due to the variance of the irradiation amount, differs depending upon the pattern drawing area ratio.
FIG. 1 shows the relationship between the pattern drawing area ratio and the amount of the variance in line width of a pattern, obtained in the case where the irradiation amount varies by 1 .mu.C/cm.sup.2. In the case where the pattern drawing area ratio is close to 100%, when the irradiation amount varies by 1 .mu.C/cm.sup.2, the line width of the resist pattern varies by about 23 nm, whereas when the drawing area ratio is close to 0%, the variance of the measurement of the line width can be limited small as about 10 nm for the same irradiation variance amount.
In the meantime, the correction amount is not in the form of continuous data, but is divided into 256 gradations, as described above. Since there are such gradations, stepwise differences, although they are small, are created in the correction amount data. For example, in the case where the correction amount for one gradation is 0.7% of the reference irradiation amount, an error of .+-.0.35% may be created. When this is applied to the resist whose reference irradiation amount 20 .mu.C/cm.sup.2, the error in the irradiation amount is .+-.0.07 .mu.C/cm.sup.2. Since the irradiation amount error is constant regardless of the drawing area ratio, in a region where the drawing area ratio is close to 0%, the error in dimension is about 0.7 nm, whereas in a region where the drawing area ratio is close to 100%, the error in dimension is about 1.6 nm. Thus, for different drawing area ratios, different effective irradiation amounts result. Therefore, the proximity effect correction efficiency differs from one case to another.
Further, usually, the optimal irradiation amount obtained when the drawing area ratio is 100% is assigned as a reference irradiation amount, and the range from a drawing area ratio of 100% to 0% is separated into 256 gradations to carry out the correction. Therefore, naturally, even in the case where the drawing area ratio is placed between 0% and 50%, the range between a drawing area ratio of 0% and that of 100% is separated into 256 gradations. Accordingly, the correction accuracy becomes poor.
In the meantime, there is recently a high demand of increasing the accuracy of photomasks used in the lithography step of the manufacture of semiconductors. More specifically, there is a demand of achieving a dimension accuracy of 10 nm (3 .sigma.) for a pattern on a photomask in the manufacture of a 0.13 .mu.m-device. There are a number of factors which causes an error in the pattern dimension of a photomask, for example, the error in the form of beam used in the electron beam drawing device, the variance in the irradiation amount, and the variation in the sensitivity of the resist. Therefore, it is required that the error in dimension which is due to the proximity effect correction should be 3 nm or less at least. There are some factors for errors caused due to the proximity effect correction, that is, for example, the error in the calculation of the proximity effect correction and the error of the corrected gradations. Here, it is preferable that the error in dimension due to the error of the corrected gradations should be 1 nm or less. Thus, the conventional proximity effect correcting method entails a problem that the error in dimension exceeds an allowable limit in the region where the drawing area ratio is large.